Engineering intelligence into real-time applications

: While significant advances have been made in real‐time expert systems, it is Often a major problem to match the software with an appropriate delivery vehicle. We describe two similar approaches, developed independently and brought together in this paper, that address both the language and hardware issues. The result is an integrated solution that preserves the benefits of high‐level languages for developing expert systems while meeting the constraints of many real‐time applications for execution speed and data asynchronism. These constraints are met without the need for translating a high‐level system into a low‐level software, description. We take as a premise that expert systems technology adds value to control and data acquisition systems, and focus on the problem of engineering this technology into systems with severe resource bounds, difficult performance constraints, and non‐standard hardware. Rather than taking the frontal approach of forcing predeveloped, knowledge‐based code into a target machine and attempting to integrate it with a real‐time component, we argue for a platform language as a basis for both the real‐time and knowledge‐based portions. We describe our development work and applications, which demonstrate that this strategy is practical and effective.